Comparative Analysis of Nyquist Window Functions in OFDM System
Keywords
Abstract
OFDM or orthogonal frequency-division multiplexing is one of the mutilcarrier systems widely used in modern telecommunications. Overlap of spectrums allows to enhance usage of limited frequency resource. The need to preserve orthogonality between subcarriers causes high sensitivity of reception to frequency offsets, e.g. the Doppler effect in case of mutually moving transmission and reception which leads to interchannel interference (ICI). The task of researching methods to reduce the level of ICI in an OFDM system is therefore relevant. ICI level is determined by the shape of signal constructions that are used in transmitter and receiver of OFDM system. At the same time choice of signal constructions is limited by their spectral densities which must have zeros at equal frequency distances. This feature is linked to Discrete Fourier Transform at the receiver side that is used to define transmitted modulated symbols in each subchannel. Rectangular window function Is used in conventional OFDM system that has a handicap in too high ICI level at different frequency offsets. In this paper new Nyquist window functions are investigated which are created by convolution with rectangular function. New expressions for the spectral densities of selected functions were obtained by the authors. New results of the analysis of ICI level when changing the parameter of transition region duration of receiver window functions were obtained. The variation of the signal-to-interference ratio (SIR) has been investigated for each Nyquist window function when changing the frequency offset value. Estimation of bit error probability (BER) for BPSK modulation in the presence of interchannel interference when employing Nyquist window functions at the receive is executed r. The obtained new research results are presented in the form of graphs constructed in the Matlab environment. Based on the analysis results, it was found that the trapezoidal-like pulse has the highest SIR level and the lowest BER level among the investigated pulses, which is associated with the lowest level of the first two side lobes of its spectral density.
